|
Home > Archive > BizTalk Server General > November 2004 > Processor Usage
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
|
|
|
| How to make BizTalk use 100% processor? I am working on an
application on a Intel Xeon with 4 processors and 8GB RAM.
All I see is, a processor utilization of 30% and a memory
usage of 3GB max. The application takes a while to run.
How to tweak Biztalk configuration to make it use most of
the processing power and complete the job soon?
Thanks,
Rick
| |
| Todd_Uhl 2004-11-22, 8:46 pm |
| Hmmm...a slightly odd request, but I see where you're headed ;-). You have a
pretty beefy machine, so maxing it out may prove tough. I would suggest you
read through the performance characteristics document which you can find on
the BPI site. Are you dealing with a large file, or trying to process
multiple files at a time? Is your throughput lower than you were hoping, and
you thought being that using more of the processor will increase throughput?
Todd Uhl
http://www.geekswithblogs.net/toddu
"Rick" wrote:
> How to make BizTalk use 100% processor? I am working on an
> application on a Intel Xeon with 4 processors and 8GB RAM.
> All I see is, a processor utilization of 30% and a memory
> usage of 3GB max. The application takes a while to run.
> How to tweak Biztalk configuration to make it use most of
> the processing power and complete the job soon?
>
> Thanks,
> Rick
>
| |
|
| Todd,
Thank you for your reply. As you assumed, I have a beefy
machine and want to utilise all of the processing
capability to improve throughput and reduce the latency. I
am involved in testing optimal file size for our
application. So I am sending huge files as well as lots of
small files. How to make my application use all the
processing power and memory?
I was about to try to tweak the paramaeters in
adm_serviceClass table. But I am not quite clear about the
whole thing. For example the WaterMark for Orchestrations.
By default it is between 10-20. But some times the number
of Orchestration instances are about 100. They reduce to
some thing between 0-20 (Mostly below 10) before going up
again. But as per documentation when they exceed High mark
they will be brought down to something below Low Mark
before creating new instances. But I found it is not the
case always. New orchestrations are created when there are
11-12 orchestrations in executing which is obviously not
less than low mark. And what factor determines the number
of Orchestration created at a time. i.e. why 100 instances
and why not 1000? I dropped 20000 files. Even then the
maximum never crossed 100 at any time.
Sorry for huge mail. I would appreciate any help
clarifying these to me.
Thank you
>-----Original Message-----
>Hmmm...a slightly odd request, but I see where you're
headed ;-). You have a
> pretty beefy machine, so maxing it out may prove tough.
I would suggest you
>read through the performance characteristics document
which you can find on
>the BPI site. Are you dealing with a large file, or
trying to process
>multiple files at a time? Is your throughput lower than
you were hoping, and
>you thought being that using more of the processor will
increase throughput?
>
>Todd Uhl
>http://www.geekswithblogs.net/toddu
>
>"Rick" wrote:
>
an[vbcol=seagreen]
RAM.[vbcol=seagreen]
memory[vbcol=seagreen]
of[vbcol=seagreen]
>.
>
| |
|
| Thumbs down to BizTalk. When it cant use or cant atleast
show how to maximize the utilization of all the processing
capability then the product is not suitable for Enterprise
applications.
For all those who point to BizTalk Performance document, I
already read it 1000 times. I advise you guys to read once
again. There is nothing specific in that document which
gives specifics about performance tuning. Its just a
summary of some tests done by MS. Even though it mentions
about some configurable options, there is nothing concrete
which shows how their tests performed for specific
configuration.
Rick
>-----Original Message-----
>Todd,
>
>Thank you for your reply. As you assumed, I have a beefy
>machine and want to utilise all of the processing
>capability to improve throughput and reduce the latency.
I
>am involved in testing optimal file size for our
>application. So I am sending huge files as well as lots
of
>small files. How to make my application use all the
>processing power and memory?
>
>I was about to try to tweak the paramaeters in
>adm_serviceClass table. But I am not quite clear about
the
>whole thing. For example the WaterMark for
Orchestrations.
>By default it is between 10-20. But some times the number
>of Orchestration instances are about 100. They reduce to
>some thing between 0-20 (Mostly below 10) before going up
>again. But as per documentation when they exceed High
mark
>they will be brought down to something below Low Mark
>before creating new instances. But I found it is not the
>case always. New orchestrations are created when there
are
>11-12 orchestrations in executing which is obviously not
>less than low mark. And what factor determines the number
>of Orchestration created at a time. i.e. why 100
instances
>and why not 1000? I dropped 20000 files. Even then the
>maximum never crossed 100 at any time.
>
>Sorry for huge mail. I would appreciate any help
>clarifying these to me.
>
>Thank you
>
>
>
>headed ;-). You have a
tough.[vbcol=seagreen]
>I would suggest you
>which you can find on
>trying to process
>you were hoping, and
>increase throughput?
on[vbcol=seagreen]
>an
>RAM.
>memory
run.[vbcol=seagreen]
>of
>.
>
|
|
|
|
|